Bug 116366

Created attachment 49530 [details]
yast2 and a few other logs

Created attachment 49531 [details]
diff between 9.3 and 10.0RC1 xorg.conf

The attached logs are taken after switching to runlevel 3, and trying to run 
sax2, which also barfed and refuses to start. 
 
This is basically a k.o.-bug for a non-tech-savvy person. 

Can't get sax2 to start, not with any combination of -r and -l. 
 
Deleted the xorg-* and 3 other related packages and installed the equivalent 
packages from 9.3, copied xorg.conf from 9.3. 10.0RC1 still doesn't start the 
X server. Booting back into 9.3, all runs fine at 1024x768, 70Hz, 56kHz. 
 
No graphics on this machine when it ought to work. Changing severity to major. 

Monitor detection and modeline generation differs from 9.3. SaX2 specific. 
 
-  HorizSync    28-56 
+  HorizSync    28-48 
[...] 
-  VertRefresh  50-70 
+  VertRefresh  50-60 
[...] 
+  Modeline 	"1024x768" 61.89 1024 1080 1184 1344 768 769 772 794 
-  Modeline 	"1024x768" 75.00 1024 1048 1184 1328 768 771 777 806 
+  Modeline 	"1024x768" 65.0 1024 1048 1184 1344 768 771 777 806 -hsync 
-vsync 

Well, now it suddenly works - X starts fine. Changes to the system: packages  
glibc-devel-2.3.5-39, gcc-4.0.2_20050901-3, kernel-source-2.6.13-8 installed  
to compile conexant driver. Packages were installed with yast, ie SuSEconfig  
was run. When X didn't start, I had just finished installing a default KDE  
system, though I did add a few packages and removed some others (games, cups,  
isdn, scanning). All dependencies were resolved when installation started.  
  
This is still with the xorg from 9.3. When I remove those 9,3 packages and use  
yast to reinstall xorg from 10.0RC1, and run init3; init 5 the X server still  
works. xorg.conf is still copied from 9.3 but this isn't the relevant factor.  
  
Removing kernel-source, gcc, glbic-devel, SuSEconfig; init 3; init 5 doesn't  
cause an X server startup failure any longer.  
  
I can't track the cause, but the lack of something one of the packages  
contains or does at first boot causes an X server crash. I have no more ideas.  

Re comment #5: the first thing I had done was copy the working xorg.conf from 
9.3 over the one of 10.0RC1. Absolutely no difference. Also, sax2 should be 
able to start up regardless how screwy xorg.conf might be - otherwise I'll 
never get a working system! 

The log told me: 
 
----snip------ 
 
(==) I810(0): Write-combining range (0xd8000000,0x4000000) 
    (II) I810(0): vgaHWGetIOBase: hwp->IOBase is 0x03d0, hwp->PIOOffset is 
0x0000    
    pgetbl_ctl: 0xef3de001 pgetbl_err: 0x39 
    ipeir: 0 iphdr: 0 
    LP ring tail: 1ace98 head: 0 len: 1ac003 start 27ac000 
    eir: 0 esr: 10 emr: 38 
    instdone: ff7a instpm: 1d 
    memmode: 5 instps: c10 
    hwstam: 8a45 ier: 8a45 imr: 8a45 iir: 0 
 
    Fatal server error: 
    Active ring not flushed 
 
----snap---- 
 
This is not a SaX bug 

Ouch! Matthias, we need to test our i815 machine ... 

I remember that we had these 'Active ring not flushed' errors some time ago,
they appeared on my i815 system as well, but ought to be fixed.
As soon as this 'Active ring not flushed' message appears once, the Xserver will
never ever start up again until the system is rebooted.

Everything worked on my i815 as anticipated. I restarted the X server 10 times,
no problems.

Closing this as WORKSFORME, please reopen if the error occurs again. Maybe this
was just some alpha particle from outer space...

Thanks for testing!